U.S. patent application number 14/460500 was filed with the patent office on 2015-02-19 for quantization method, image processing apparatus, and recording medium.
This patent application is currently assigned to FUJIFILM Corporation. The applicant listed for this patent is FUJIFILM Corporation. Invention is credited to Hiroyuki SHIBATA.
Application Number | 20150049366 14/460500 |
Document ID | / |
Family ID | 52466642 |
Filed Date | 2015-02-19 |
United States Patent
Application |
20150049366 |
Kind Code |
A1 |
SHIBATA; Hiroyuki |
February 19, 2015 |
QUANTIZATION METHOD, IMAGE PROCESSING APPARATUS, AND RECORDING
MEDIUM
Abstract
A quantization method according to an aspect of the present
invention includes the steps of quantizing a first image data by
the use of a basic pattern and converting the first image data into
a second image data that represents a binary or multi-level
quantized pattern having a gray level smaller than that of the
first image data. The basic pattern according to this aspect of the
present invention presents high frequent occurrence of the basic
tone patterns and a mostly uniform-distributed pattern of the
clusters of different kinds of the basic tone patterns in the image
with the long-distance autocorrelation (periodicity) of the basic
tone patterns suppressed. Quantization by the use of this basic
pattern provides a quantized pattern that reflects pattern
characteristics of the basic pattern.
Inventors: |
SHIBATA; Hiroyuki;
(Ashigarakami-gun, JP) |
|
Applicant: |
Name |
City |
State |
Country |
Type |
FUJIFILM Corporation |
Tokyo |
|
JP |
|
|
Assignee: |
FUJIFILM Corporation
Tokyo
JP
|
Family ID: |
52466642 |
Appl. No.: |
14/460500 |
Filed: |
August 15, 2014 |
Current U.S.
Class: |
358/3.07 |
Current CPC
Class: |
H04N 1/40 20130101; H04N
1/4051 20130101 |
Class at
Publication: |
358/3.07 |
International
Class: |
G06K 15/10 20060101
G06K015/10; G06K 15/02 20060101 G06K015/02 |
Foreign Application Data
Date |
Code |
Application Number |
Aug 19, 2013 |
JP |
2013-169826 |
Claims
1. A quantization method, comprising the steps of: quantizing a
first image data by the use of a basic pattern; and converting the
first image data into a second image data that represents a binary
or multi-level quantized pattern having a gray level smaller than
that of the first image data, wherein, when each of basic tone
frequencies is a local maximum of spatial frequency components in a
pattern image that contains each of basic tone patterns repeatedly
arranged as a repeating unit of a specific pattern in a
two-dimensional dot arrangement, the basic pattern has spatial
frequency characteristics in which components at and in the
neighborhood of each basic tone frequency are relatively suppressed
in comparison with other spatial frequency components in the
pattern image, the local maximum exists in components at the outer
periphery of each basic tone frequency outside the neighborhood of
each basic tone frequency, and components at low frequencies are
suppressed.
2. The quantization method according to claim 1, wherein the basic
pattern is created through the steps of: making a first pattern in
which components in the neighborhood of each of the basic tone
frequencies are relatively suppressed in comparison with other
components and the local maximum of components exists at the outer
periphery of each basic tone frequency; and suppressing lower
frequency components than those of the first pattern while
maintaining pattern characteristics of the neighborhood and the
outer periphery of each basic tone frequency in the first
pattern.
3. The quantization method according to claim 2, wherein the step
of making the first pattern includes the steps of: making a
division pattern divided into N different regions (N: 2 or
greater), the division pattern having pattern characteristics in
which, out of spatial frequency components, low-frequency
components lower than a first frequency Fmin and high-frequency
components higher than a second frequency Fmax which is higher than
the first frequency Fmin are suppressed; and performing convolution
of the respective N regions in the division pattern with N
different basic tone patterns having identical densities per unit
area and having a difference in at least one of phase and basic
tone frequency, one to the other, wherein the step of suppressing
lower frequency components than those of the first pattern includes
the step of converting the first pattern made in the convolution
step into a second pattern in which respective basic tone frequency
components and respective low-frequency components in the N
different basic tone patterns are suppressed.
4. The quantization method according to claim 3, wherein the step
of making the division pattern includes the steps of: applying a
band-pass filter for suppressing low-frequency components lower
than the first frequency Fmin and high-frequency components higher
than the second frequency Fmax to a white noise pattern; and
applying N-1 threshold levels to a pattern resulting from the
application of the band-pass filter so as to divide the pattern
into the N regions.
5. The quantization method according to claim 2, wherein the step
of suppressing lower frequency components than those of the first
pattern includes the steps of: performing filter processing on a
pattern; and performing an exchange between relatively high density
dots and relatively low density non-dots in the pattern, and
wherein the filter processing uses a filter that highlights
low-frequency components and components in the neighborhood of each
basic tone frequency among other components in the pattern.
6. The quantization method according to claim 2, wherein the step
of suppressing lower frequency components than those of the first
pattern is a step for extracting boundary regions including the
boundaries of the N regions from the first pattern and changing the
arrangement of dots only in the boundary regions.
7. The quantization method according to claim 1, wherein the basic
pattern has a record ratio of 50%.
8. The quantization method according to claim 1, wherein the
quantization is processed by the use of a threshold matrix created
based on the basic pattern.
9. The quantization method according to claim 1, wherein the
quantization is processed by an error diffusion method through the
use of the basic pattern as a constraint for dot arrangement.
10. An image processing apparatus comprising: an image input part
that captures a first image data; and a quantization processing
part for quantizing the first image data and converting the first
image data into a second image data that represents a binary or
multi-level quantized pattern having a gray level smaller than that
of the first image data, wherein the quantization processing part
quantizes the first image data by the use of a basic pattern to
convert the first image data into the second image data, and when
each of basic tone frequencies is a local maximum of spatial
frequency components in a pattern image that contains each of basic
tone patterns repeatedly arranged as a repeating unit of a specific
pattern in a two-dimensional dot arrangement, the basic pattern has
spatial frequency characteristics in which components at and in the
neighborhood of each basic tone frequency are relatively suppressed
in comparison with other spatial frequency components in the
pattern image, the local maximum exists in components at the outer
periphery of each basic tone frequency outside the neighborhood of
each basic tone frequency, and components at low frequencies are
suppressed.
11. The image processing apparatus according to claim 10, further
comprising a threshold matrix storing part that stores a threshold
matrix created based on the basic pattern, wherein the quantization
processing part processes the quantization by the use of the
threshold matrix.
12. The image processing apparatus according to either claim 10,
further comprising a basic pattern storing part that stores the
basic pattern, wherein the quantization processing part processes
the quantization by an error diffusion method through the use of
the basic pattern as a constraint for dot arrangement.
13. A non-transitory recording medium in which computer-readable
code of a program which causes a computer to perform the
quantization method according to claim 1 is stored.
Description
CROSS-REFERENCE TO RELATED APPLICATIONS
[0001] The patent application claims priority under 35 U.S.C.
.sctn.119 to Japanese Patent Application No. 2013-169826, filed on
Aug. 19, 2013. Each of the above application(s) is hereby expressly
incorporated by reference, in its entirety, into the present
application.
BACKGROUND OF THE INVENTION
[0002] 1. Field of the Invention
[0003] The present invention relates to a quantization method, an
image processing apparatus, and a recording medium, and
particularly to a quantization technique for converting a
continuous-tone image into a binary or multi-level dot image.
[0004] 2. Description of the Related Art
[0005] In the field of printing, a continuous-tone image of a
printing subject (for example, an m-level image) is converted to a
binary or multi-level dot image (an n-value image) by quantization
processing such as dithering and an error diffusion method, where m
and n are integers that satisfy 2.ltoreq.n<m. As a result, an
image is formed in accordance with data on the obtained dot image,
as is disclosed in Japanese Patent No. 4143560 and "Dithering with
blue noise." by Ulichney, Robert A., Proceedings of the IEEE
76.1(1988): 56-79.
[0006] For example, in some inkjet recording apparatuses that form
color images, the variety of ink colors is increased to expand the
color reproduction region by adding light cyan (LC), light magenta
(LM) and special colors to regular ink colors, i.e. cyan (C),
magenta (M), yellow (Y), and black (Bk). While a method for
increasing the number of ink color variations for use is effective
in expanding the color reproduction region, it disadvantageously
increases the cost of the apparatus.
SUMMARY OF THE INVENTION
[0007] Another method for expanding the color reproduction region
uses a dot pattern for representing gray levels in which dots are,
to the extent possible, arranged evenly without being overlapped
one another in an image. This enables the color reproduction region
to be expanded without an increase in ink color variation.
[0008] When it is supposed, for example, that dots can be printed
on a square lattice corresponding to the pixel array of a
two-dimensional image, the most ideal pattern among dot patterns
with a record ratio of 50%, which is effective in a color
reproduction region, is a checkered pattern having a dot
arrangement in which a 2-pixel unit comprising dot-on (dot
presence) and dot-off (dot absence) are alternately repeated in
both the horizontal (x) and vertical (y) directions. Use of the
checkered pattern for quantization enables a color reproduction
region to be expanded.
[0009] Unfortunately, frequent use of checkered patterns in a dot
image tends to produce artifacts due to the following facts.
[0010] Disadvantage 1: The boundaries of regions of checkered
patterns having different phases are irregularly viewable.
[0011] Disadvantage 2: When a checkered pattern with high
periodicity (long-distance autocorrelation) is used, interference
between the printing operation of the inkjet head and the checkered
pattern leads to an increased amount of ink discharge at an
identical timing, thus generating an artifact owing to crosstalk
(mutual fluidal interference through a channel in the head).
[0012] Disadvantage 3: When a checkered pattern with high
periodicity is used, the checkered pattern can interfere with a
mechanical vibration error in the inkjet recording apparatus (for
example, mechanical vibration in a paper conveying system), causing
artifacts to be noticeable.
[0013] A blue noise pattern can resolve the artifact disadvantages
described above to some extent, as disclosed in "Dithering with
blue noise." by Ulichney, RobertA., Proceedings of the IEEE
76.1(1988): 56-79. The blue noise pattern is a pattern having a
spectrum in which white noise at a high frequency side and energy
at a low frequency side are suppressed. Since the blue noise
pattern has low periodicity on the checkered pattern part, the
disadvantages 1 to 3 described above are restrained.
[0014] Unfortunately, the blue noise pattern leads to a relatively
suppressed spectrum on the high frequency side, resulting in
not-high frequent occurrence of checkered patterns that are helpful
for expanding a color reproduction region. Thus, it is deficient in
terms of color reproduction region. In other words, the
conventional arts present a trade-off relationship between
expanding the color reproduction region and reducing artifacts.
[0015] The disadvantages described above are not only relevant to
inkjet recording apparatuses, but are understood as disadvantages
common to various image forming apparatuses that use dot recording
to represent gray levels.
[0016] It is an object of the present invention to provide a
quantization method, an image processing apparatus, and a recording
medium capable of creating a quantized pattern allowing both the
expansion of a color reproduction region and the suppression of
artifacts.
[0017] To resolve the disadvantages, the invention described below
includes:
[0018] (First aspect): A quantization method according to a first
aspect of the present invention includes the steps of quantizing a
first image data by the use of a basic pattern and converting the
first image data into a second image data that represents a binary
or multi-level quantized pattern having a gray level smaller than
that of the first image data. When each of basic tone frequencies
is a local maximum of spatial frequency components in a pattern
image that contains each of basic tone patterns repeatedly arranged
as a repeating unit of a specific pattern in a two-dimensional dot
arrangement, the basic pattern has spatial frequency
characteristics in which components at and in the neighborhood of
each basic tone frequency are relatively suppressed in comparison
with other spatial frequency components in the pattern image, the
local maximum exists in components at the outer periphery of each
basic tone frequency outside the neighborhood of each basic tone
frequency, and components at low frequencies are suppressed.
[0019] The basic pattern according to the first aspect of the
present invention presents high frequent occurrence of the basic
tone patterns and a mostly uniform-distributed pattern of the
clusters of different kinds of the basic tone patterns in the image
with the long-distance autocorrelation (periodicity) of the basic
tone patterns suppressed. Quantization by the use of this basic
pattern provides a quantized pattern that reflects pattern
characteristics of the basic pattern.
[0020] The first image data is pre-quantization data such as
continuous-tone image data of m gray levels. The second image data
is an n-level quantized image obtained through quantization of the
first image data, where m and n are integers that satisfy
2.ltoreq.n<m.
[0021] The quantization method according to the first aspect of the
present invention provides a quantized pattern that has
characteristics of the basic pattern, allowing both the expansion
of a color reproduction region and the suppression of
artifacts.
[0022] The basic pattern is defined as a pattern corresponding to
certain gray levels (halftones as standard, e.g. a record ratio of
50%). In quantization processing, a dot arrangement for the basic
pattern is implemented for gray levels corresponding to the basic
pattern. The characteristics of the basic pattern are nearly kept
in quantized patterns for other gray levels (halftones) created in
accordance with the basic pattern.
[0023] The range of the "neighborhood" in the "neighborhood of a
basic tone frequency" can be set to an appropriate range depending
on the printing resolution and the like. As an example, with 1
pixel ([px]) in the case of a printing resolution of 1200 dpi taken
as a unit, the range of a frequency difference approximately from
1/20 [cycle/px] to 1/3 [cycle/px] can be set to a range for the
neighborhood.
[0024] The "low-frequency components" denote components at low
frequency side that affect viewability with consideration given to
frequency characteristics of the human eye. As a guideline,
components in a low-frequency range of about 10 cycle/mm or below
are equivalent to this.
[0025] The "use of a basic pattern" is not limited to the direct
use of the dot arrangement of a basic pattern but includes modes
where a basic pattern is indirectly used, such as in cases where
quantization is processed through the use of a threshold matrix
made based on a basic pattern.
[0026] (Second aspect): In the quantization method according to the
first aspect of the present invention, the basic pattern is created
through the steps of: making a first pattern in which components in
the neighborhood of each of the basic tone frequencies are
relatively suppressed in comparison with other components and the
local maximum of components exists at the outer periphery of each
basic tone frequency; and suppressing lower frequency components
than those of the first pattern while maintaining pattern
characteristics of the neighborhood and the outer periphery of each
basic tone frequency in the first pattern.
[0027] A method for creating the basic pattern is composed of a
process which includes the steps of: making a first pattern in
which components in the neighborhood of each of the basic tone
frequencies are relatively suppressed in comparison with other
components and the local maximum of components exists at the outer
periphery of each basic tone frequency; and suppressing lower
frequency components than those of the first pattern while
maintaining pattern characteristics of the neighborhood and the
outer periphery of each basic tone frequency in the first
pattern.
[0028] (Third aspect): In the quantization method according to the
second aspect of the present invention, the step of making the
first pattern includes the step of making a division pattern
divided into N different regions (N: 2 or greater). The division
pattern has pattern characteristics in which, out of spatial
frequency components, low-frequency components lower than a first
frequency Fmin and high-frequency components higher than a second
frequency Fmax which is higher than the first frequency Fmin are
suppressed. The step of making the first pattern further includes
the step of performing convolution of the respective N regions in
the division pattern with N different basic tone patterns having
identical densities per unit area and having a difference in at
least one of phase and basic tone frequency, one to the other. The
step of suppressing lower frequency components than those of the
first pattern includes the step of converting the first pattern
made in the convolution step into a second pattern in which
respective basic tone frequency components and respective
low-frequency components in the N different basic tone patterns are
suppressed.
[0029] (Fourth aspect): In the quantization method according to the
third aspect of the present invention, the step of making a
division pattern includes the steps of: applying a band-pass filter
for suppressing low-frequency components lower than the first
frequency Fmin and high-frequency components higher than the second
frequency Fmax to a white noise pattern; and applying N-1 threshold
levels to a pattern resulting from the application of the band-pass
filter so as to divide the pattern into the N regions.
[0030] (Fifth aspect): In the quantization method according to any
one of the second to fourth aspects of the present invention, the
step of suppressing lower frequency components than those of the
first pattern includes the steps of: performing filter processing
on a pattern; and performing an exchange between relatively high
density dots and relatively low density non-dots in the pattern.
The filter processing uses a filter that highlights low-frequency
components and components in the neighborhood of each basic tone
frequency in comparison with other components in the pattern.
[0031] (Sixth aspect): In the quantization method according to any
one of the second to fourth aspects of the present invention, the
step of suppressing lower frequency components than those of the
first pattern is a step for extracting boundary regions including
the boundaries of the N regions from the first pattern and changing
the arrangement of dots only in the boundary regions.
[0032] Correcting the dot arrangement only within the boundary
regions enables the suppression of low-frequency components while
maintaining pattern characteristics in the neighborhood of each of
the basic tone frequencies.
[0033] (Seventh aspect): In the quantization method according to
any one of the first to sixth aspects of the present invention, the
basic pattern has a record ratio of 50%.
[0034] It is preferable that the basic pattern should have a record
ratio of 50% which produces the most profound effect on the
expansion of the color reproduction region in the pattern. The
record ratio, however, is not limited to 50% in applying the
invention. It may be around 50% of halftone. The allowable range of
around 50% is, for example, 50%.+-.about 10%.
[0035] (Eighth aspect): In the quantization method according to any
one of the first to seventh aspects of the present invention, the
quantization is processed by the use of a threshold matrix created
based on the basic pattern.
[0036] (Ninth aspect): In the quantization method according to any
one of the first to eighth aspects of the present invention, the
quantization is processed by an error diffusion method through the
use of the basic pattern as a constraint for dot arrangement.
[0037] A mode of jointly using the threshold matrix described in
the eighth aspect and the error diffusion method described in the
ninth aspect for quantization can be employed.
[0038] (Tenth aspect): An image processing apparatus according to a
tenth aspect of the present invention includes: an image input part
that captures a first image data; and a quantization processing
part for quantizing the first image data and converting the first
image data into a second image data that represents a binary or
multi-level quantized pattern having a gray level smaller than that
of the first image data. The quantization processing part quantizes
the first image data by the use of a basic pattern to convert the
first image data into the second image data. When each of basic
tone frequencies is a local maximum of spatial frequency components
in a pattern image that contains each of basic tone patterns
repeatedly arranged as a repeating unit of a specific pattern in a
two-dimensional dot arrangement, the basic pattern has spatial
frequency characteristics in which components at and in the
neighborhood of each basic tone frequency are relatively suppressed
in comparison with other spatial frequency components in the
pattern image, the local maximum exists in components at the outer
periphery of each basic tone frequency outside the neighborhood of
each basic tone frequency, and components at low frequencies are
suppressed.
[0039] The image processing apparatus according to the tenth aspect
of the present invention quantizes a pre-quantization first image
data to convert the image data into a post-quantization second
image data (quantized pattern). Quantized patterns obtained by an
image processing apparatus of this mode present high frequent
occurrence of the basic tone patterns and a mostly
uniform-distributed pattern of the clusters of different kinds of
the basic tone patterns in the image, allowing both the expansion
of the color reproduction region and the suppression of
artifacts.
[0040] The image processing apparatus according to the tenth aspect
of the present invention can be appropriately combined with
particulars similar to those on the "basic pattern" described in
the second to seventh aspects.
[0041] (Eleventh aspect): The image processing apparatus according
to the tenth aspect of the present invention further includes a
threshold matrix storing part that stores a threshold matrix
created based on the basic pattern. The quantization processing
part processes quantization by the use of the threshold matrix.
[0042] (Twelfth aspect): The image processing apparatus according
to either the tenth aspect or the eleventh aspect of the present
invention further includes a basic pattern storing part that stores
the basic pattern. The quantization processing part processes
quantization by an error diffusion method through the use of the
basic pattern as a constraint for dot arrangement.
[0043] (Thirteenth aspect): A non-transitory recording medium
according to a thirteenth aspect of the present invention stores
computer-readable code of a program which causes a computer to
perform the quantization method according to any one of the first
to the ninth aspect.
[0044] According to the present invention, a high-quality,
favorable dot image can be created, allowing both the expansion of
the color reproduction region and the suppression of artifacts.
BRIEF DESCRIPTION OF THE DRAWINGS
[0045] FIG. 1 illustrates checkered patterns having different
phases;
[0046] FIG. 2 illustrates an example of a preferable pattern
created in accordance with an embodiment of the present
invention;
[0047] FIG. 3 illustrates an example of a pattern to show
unresolved disadvantages in comparison;
[0048] FIG. 4 is a schematic diagram illustrating frequency
characteristics (two-dimensional spatial frequency spectrum) of a
preferable pattern;
[0049] FIG. 5 is a schematic diagram of ideal (desirable) spatial
frequency characteristics;
[0050] FIG. 6 illustrates spatial frequency characteristics of a
pattern obtained in accordance with an embodiment of the present
invention;
[0051] FIG. 7 illustrates spatial frequency characteristics of a
pattern in accordance with the comparative example;
[0052] FIG. 8 is a flowchart showing a method of creating a basic
pattern in accordance with an embodiment of the present
invention;
[0053] FIG. 9 illustrates an example of a pattern D(x) divided in N
(N=2) regions;
[0054] FIG. 10 illustrates spatial frequency characteristics of the
pattern D(x) divided in N regions;
[0055] FIG. 11 is an enlarged view of a part of upper left regions
in FIG. 9;
[0056] FIG. 12 is a flowchart showing an organized process for
creating a first pattern;
[0057] FIGS. 13A, 13B and 13C are an example of a first pattern
created by processing in steps A-1 and A-2, an enlarged view of a
part of an upper left region of FIG. 13A, and a diagram
illustrating frequency characteristics (two-dimensional spatial
frequency spectrum) of the pattern of FIG. 13A, respectively;
[0058] FIGS. 14A, 14B and 14C are an example of a pattern
(equivalent to a basic pattern) ultimately created by processing in
a process B, an enlarged view of a part of an upper left region of
FIG. 14A, and a diagram illustrating frequency characteristics
(two-dimensional spatial frequency spectrum) of the pattern of FIG.
14A, respectively;
[0059] FIG. 15 is a flowchart showing an example of dot exchange
processing applied in an example 1;
[0060] FIG. 16 is an explanatory drawing of frequency
characteristics of a dot selection filter;
[0061] FIG. 17 is a graph showing filter intensity of a dot
selection filter in a radial direction;
[0062] FIG. 18 shows a concrete example of a Point Spread Function
(PSF) filter;
[0063] FIG. 19 is a flowchart showing dot exchange processing
applied in an example 2;
[0064] FIG. 20 is an enlarged view illustrating clusters of
checkered patterns and their boundaries in an easy-to-understand
manner;
[0065] FIG. 21 is a diagram showing an organized overview of a
method for creating a pattern in accordance with an embodiment of
the present invention;
[0066] FIG. 22 is a block diagram illustrating a configuration
example of an apparatus for creating a pattern in accordance with
an embodiment of the present invention;
[0067] FIG. 23 is a block diagram illustrating a configuration
example of an image processing apparatus in accordance with an
embodiment of the present invention;
[0068] FIG. 24 is a block diagram illustrating a configuration
example of an image processing apparatus in accordance with another
embodiment of the present invention;
[0069] FIG. 25 shows an example of a matrix for error
diffusion;
[0070] FIG. 26 is a flowchart showing an example of quantization
processing in which a threshold matrix and an error diffusion
method are jointly used;
[0071] FIG. 27 is a block diagram illustrating a configuration
example of an image processing apparatus in accordance with another
embodiment of the present invention;
[0072] FIG. 28 illustrates other examples of a basic tone
pattern;
[0073] FIG. 29 illustrates a checkered pattern image in a real
space;
[0074] FIG. 30 illustrates an image obtained when a fast Fourier
transform (FFT) is performed on the image of FIG. 29;
[0075] FIG. 31 illustrates a 1- by 2-pixel checkered pattern image
in a real space;
[0076] FIG. 32 illustrates an image obtained when a FFT is
performed on the image of FIG. 31; and
[0077] FIG. 33 is a block diagram illustrating essential components
of an inkjet recording apparatus.
DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
[0078] Embodiments of the present invention will now be described
in detail with reference to the attached drawings.
[0079] <Organizing Disadvantages and Description of Solution
Principles>
[0080] To make explanation simple in this description, let
"checkered patterns" be taken as an example of basic tone patterns
which have spatial periodicity and produce effect on the expansion
of a color reproduction region. As described earlier, disadvantages
associated with high frequent use of checkered patterns in an image
are artifacts. When the disadvantages 1 to 3 described earlier are
perceived as disadvantages and countermeasures against them in a
pattern of dot arrangement, they are organized as follows:
[0081] Disadvantage A: The disadvantage 1 can be perceived that as
the boundaries (or clusters) of distributed checkered patterns have
low-frequency components, the boundaries (contours of clusters) are
visually identified with ease.
[0082] Countermeasure A: As a countermeasure against the
disadvantage A, spatial frequency components in the boundaries (or
clusters) of checkered patterns need to be suppressed.
[0083] Disadvantage B: The disadvantages 2 and 3 can be perceived
that the increased frequency of occurrence of checkered patterns
causes the periodicity to be facilitated, thus generating artifacts
in a wide area.
[0084] The "facilitated periodicity" means that the long-distance
autocorrelation of the pattern is enhanced. In other words,
increasing the frequency of occurrence of checkered patterns
facilitates the long-distance autocorrelation of the pattern. When
an error factor (e.g. mechanical vibration in an apparatus and
crosstalk associated with simultaneous discharge by a plurality of
nozzles of an inkjet head) observed in a long-distance range
occurs, artifacts occur in a wide area.
[0085] Countermeasure B: As a countermeasure against the
disadvantage B, the periodicity (long-distance autocorrelation) of
the pattern or the boundaries of checkered patterns (or clusters of
checkered patterns) needs to be suppressed.
[0086] In other words, the disadvantages A and B can be resolved by
controlling the boundaries of checkered patterns (or clusters of
checkered patterns) (countermeasures A, B) in addition to
suppressing low-frequency components in the pattern, which is
generally required, in quantization for converting a
continuous-tone image to a dot pattern image.
[0087] <Phases of Checkered Patterns>
[0088] FIG. 1 illustrates checkered patterns having different
phases. The checkered pattern is a pattern in which dot-on (dot
presence) and dot-off (dot absence) are alternately repeated in
both the horizontal (x) and vertical (y) directions of a
two-dimensional image lattice.
[0089] In FIG. 1, each cell of an image lattice represented by a
square lattice corresponds to a "pixel", and a white pixel
represents a non-printing pixel (dot absence) and a black pixel
represents a printing pixel (dot presence) where a dot is placed.
In some cases, a white colored image portion (pixel) without dot
(off) is called an "off portion" and a black colored image portion
(pixel) with dot (on) is called an "on portion".
[0090] Each of checkered patterns 1A, 1B in FIG. 1 is a pattern
with a record ratio of 50%. The record ratio is the percentage of
the number of printing pixels (dots) with respect to the total
number of pixels per unit area. The record ratio 100% means that a
dot is placed (recorded) on every pixel and the record ratio 50%
means that dots are placed on n/2 pixels in n pixels. FIG. 1
exemplifies an image area of 8 by 8 pixels and a dot is placed on
each of 32 pixels in 64 pixels (the number of printing dots is 32).
Although FIG. 1 shows an area of 8 by 8 pixels, a basic minimum
repeating unit for the checkered patterns 1A, 1B is 2 by 2
pixels.
[0091] White (off) portions and black (on) portions shown in the
checkered pattern 1A on the left-hand side of FIG. 1 are inversely
arranged in the checkered pattern 1B on the right-hand side, and
thus their phases of spatial repetition cycles of dot
presence/absence are different. The checkered pattern includes two
patterns having different phases as designated by reference
characters 1A and 1B.
[0092] For the convenience of description, this specification calls
the checkered pattern 1A on the left of FIG. 1 as the checkered
pattern with the "phase 0" and the checkered pattern 1B on the
right of FIG. 1 as the checkered pattern with the "phase 1".
Preferred Pattern Example
[0093] FIG. 2 illustrates an example of a preferable dot pattern
created in accordance with the embodiment. The dot pattern shown
here has a record ratio of 50%. FIG. 3, on the other hand,
illustrates an example of a pattern with the disadvantages A, B
unresolved for the sake of comparison.
[0094] In FIGS. 2 and 3, the checkered patterns 1A, 1B having
different phases described in FIG. 1 are presented with the
lightness of respective colors, gray and black, for the convenience
of description. In FIGS. 2 and 3, every region with the pattern
having dot-on portions presented in gray corresponds to the
checkered pattern 1A with "phase 0" in FIG. 1 and every region with
the pattern having dot-on portions presented in black corresponds
to the checkered pattern 1B with "phase 1" in FIG. 1.
[0095] A pattern, if it is divided into regions according to a
distinction in the phase of each checkered pattern, presents the
pattern as shown in FIG. 2 or 3. For the convenience of
description, this specification calls a closed region that consists
of checkered patterns with an identical phase as a "cluster of the
checkered pattern" and the boundary (outline) of the clusters of
the checkered patterns as a "boundary of the checkered
pattern".
[0096] The dot pattern of FIG. 2 according to the embodiment
presents suppressed low-frequency components in the pattern, high
frequent occurrence of the checkered patterns 1A, 1B and a mostly
uniform mixture of the two checkered patterns having different
phases.
[0097] In contrast, the dot pattern shown in FIG. 3 as a
comparative example presents an uneven mixture of the checkered
patterns having different phases as compared with that of FIG. 2
although it presents suppressed frequency components in the pattern
and high frequent occurrence of the checkered patterns.
[0098] <Frequency Characteristics of Pattern>
[0099] When the boundaries and clusters of the checkered patterns
as shown in FIG. 2 are represented by spatial frequency
characteristics, these are equivalent to components near the basic
tone frequency component of the checkered patterns (dashed-line
circles 5A, 5B, 5C, 5D of FIG. 4) and the volume of the "components
near the basic tone frequency component" is equivalent to the
frequency of occurrence of the checkered patterns.
[0100] FIG. 4 is a schematic diagram illustrating frequency
characteristics (two-dimensional spatial frequency spectrum). In
FIG. 4, the horizontal axis represents frequency in a paper feeding
direction and the vertical axis represents frequency in a direction
perpendicular to the paper feeding direction. The inner area of the
dashed-line circle 6 (inside the circle) with its center at the
origin point of FIG. 4 corresponds to low frequency components of
the pattern.
[0101] The four corners of FIG. 4, i.e. the centers of the
dashed-line circles 5A to 5D are equivalent to the basic tone
frequency (1/2 [cycle/px]) of the checkered patterns. As described
in the explanation of FIG. 1, the shortest spatial period of the
checkered patterns 1A, 1B is 1 period per 2 pixels ([px]), i.e. the
frequency is 1/2=0.5 [cycle/px]. This is the largest frequency and
equivalent to a "basic tone frequency".
[0102] Areas 7A to 7D around the centers in the respective
dashed-line circles 5A to 5D correspond to low-frequency components
of the clusters of the checkered patterns. The areas 7A to 7D
around the centers in the respective dashed-line circles 5A to 5D
and components of the outer peripheries, that is areas 8A to 8D
around the dashed lines correspond to the frequency of occurrence
of the checkered patterns.
[0103] Thus, suppressing the areas 7A to 7D around the centers in
the dashed-line circles 5A to 5D (low-frequency components of the
clusters of the checkered patterns) and increasing the components
of the areas 8A to 8D around the dashed lines of the dashed-line
circles 5A to 5D by increasing the frequency of occurrence of the
checkered patterns and further suppressing the inner area of the
dashed-line circle 6 of FIG. 4 (low frequency components in the
pattern) provide a solution to the disadvantages A, B.
[0104] Each of the areas 7A to 7D around the centers in the
dashed-line circles 5A to 5D includes the basic tone frequency and
constitutes a neighborhood closest to the basic tone frequency (the
most immediate neighborhood of basic tone frequency) and thus
corresponds to a "neighborhood of basic tone frequency". Each of
the components of the areas 8A to 8D around the dashed lines of the
dashed-line circles 5A to 5D corresponds to an "outer periphery of
basic tone frequency outside the neighborhood of the basic tone
frequency".
[0105] FIG. 5 is a schematic diagram of ideal (desirable) spatial
frequency characteristics. In FIG. 5, component amounts are
represented by density (brightness), thus a brighter portion
(higher intensity) has a larger component amount and a darker
portion (lower intensity) has a smaller component amount. Black
represents no component value ("component=0") and white represents
an adequately large value (near the local maximum).
[0106] In a desirable spatial frequency spectrum as shown in FIG.
5, low-frequency components in the pattern (components in an area
inside the circle designated by reference numeral 6) and
low-frequency components of the clusters of the checkered patterns
(components in the areas around the four corners designated by
reference numerals 7A to 7D, referred to as "neighborhood of basic
tone frequency") are relatively suppressed as compared with other
components and the frequency of occurrence of the checkered
patterns (components in the areas designated by reference numerals
8A to 8D, referred to as "outer periphery of basic tone frequency"
or in the areas designated by reference numerals 7A to 7D,
"neighborhood of basic tone frequency") is increased, that is,
components in the outer peripheries of each basic tone frequency
have the local maximum. Preferably, a component amount in a shaded
area 9 of FIG. 5 be close to "0" (none) although it is arbitrary to
some extent.
[0107] FIG. 6 illustrates spatial frequency characteristics of a
pattern obtained in accordance with the embodiment. FIG. 6 shows a
two-dimensional spatial frequency spectrum of the pattern shown in
FIG. 2 and the component amounts are represented by density. Rules
for density display are the same as in FIG. 5. FIG. 7 illustrates
spatial frequency characteristics of a pattern in accordance with
the comparative example. FIG. 7 shows a two-dimensional spatial
frequency spectrum of the pattern exemplified in FIG. 3.
[0108] Frequency characteristics of the pattern shown in FIG. 6
according to the embodiment meet the conditions of characteristics
described in FIG. 5. Frequency characteristics of the pattern shown
in FIG. 7 according to the comparative example, on the other hand,
do not meet the conditions because components in each area closest
to the basic tone frequency (the areas designated by reference
numerals 7A to 7D described in FIG. 5) are not suppressed although
they present a high frequency of occurrence of the checkered
patterns due to high intensity in the neighborhood of each basic
tone frequency. This results in an uneven distribution of the
clusters of the checkered patterns (refer to FIG. 3) and provides
no solution to the disadvantages A, B. In other words, artifacts
occur despite favorable color reproducibility.
[0109] <Method of Creating Pattern>
[0110] This specification calls a pattern having the desirable
frequency characteristics described in FIGS. 5 and 6 as a "basic
pattern". A method for creating the basic pattern will now be
described.
[0111] FIG. 8 is a flowchart showing a method of creating a basic
pattern in accordance with the embodiment. The method of creating a
basic pattern according to the embodiment includes the following
processes (process A and process B).
[0112] (Process A) Process for creating a pattern ("first pattern")
in which components in the neighborhood of each basic tone
frequency out of spatial frequency components in the pattern are
suppressed and the frequency of occurrence (i.e. intensity of
components) in the periphery (outer periphery of each basic tone
frequency) is increased (step S10 in FIG. 8).
[0113] In other words, the process A is a process for creating the
characteristics of four corners in the desirable frequency
characteristics diagram described in FIG. 5.
[0114] (Process B) Process for suppressing low-frequency components
while maintaining characteristics obtained in the step A (step S10
in FIG. 8) described above (step S20 in FIG. 8).
[0115] In other words, the process B is a process for creating the
characteristics of the neighborhood of the original point (central
portion) in the desirable frequency characteristics diagram
described in FIG. 5.
[0116] Through the use of a basic pattern created in the processes
A and B (steps S10, S20 in FIG. 8) for quantization processing, a
dot image that enables both the expansion of the color reproduction
region and the suppression of artifacts can be obtained.
[0117] The process A and the process B will now be described in
further detail.
[0118] <<Process a (Step S10 in FIG. 8)>>
[0119] A first pattern in which components in the neighborhood of
each basic tone frequency are suppressed and the frequency of
occurrence in the outer periphery of each basic tone frequency is
increased can be created in the following process.
[0120] (Process A-1) A pattern where low-frequency components and
high-frequency components are suppressed is converted by N-value
processing so as to divide the pattern into N regions, where N is
the number of phase types (the number of phases) of the basic tone
pattern for use and represents an integer of 2 or greater. When the
two checkered patterns 1A, 1B having different phases described in
FIG. 1 are used, the number of phases N of the basic tone pattern
for use is 2.
[0121] "A pattern where low-frequency components and high-frequency
components are suppressed" is a dot pattern having mainly green
noise characteristics (band-pass characteristics).
[0122] To take a concrete example, let us perform a convolution of
a band-pass filter (filter where low-frequency components and
high-frequency components are suppressed) with a white noise
pattern and apply N-1 threshold levels to the pattern to divide it
to N regions. To use the two checkered patterns 1A, 1B having
different phases (FIG. 1), apply 1 threshold level to the pattern
to divide it to 2 regions.
[0123] A white noise pattern can be created by specifying a size of
the pattern and generating pseudo-random numbers. [0124] white
noise (WN)=rand (pattern size) [0125] band-pass filter example
(BPF(f)); [0126] If(BPFmin<f(=frequency)<BPFmax) then,
BPF(f)=1; else, BPF(f)=0; end
[0127] Note that the cluster size of the checkered pattern depends
on BPFmin and BPFmax.
[0128] Although optimum values for BPFmin and BPFmax differ
depending on printing resolution, meeting, for example, the
following conditions is appropriate.
[0129] Fmax<1/3 cycle/px
[0130] 1/20<Fmin<1/3 cycle/px
[0131] That is to say, setting the values so that the cluster size
of the checkered patterns become larger than 3 px and up to about
20 px is proper. [0132] The following equation can be used for
convolution operation.
[0132] d(x)=ifft(fft(WN)BPF(f));
[0133] where fft represents a fast Fourier transform function.
[0134] ifft represents an inverse fast Fourier transform function.
[0135] Division into N regions (exemplified in the case of N=2)
through comparison with threshold levels can be performed by the
following computation.
[0136] N (N=2) regions division pattern D(x):
[0137] If(d(x)>th) then, D(x)=1; else D(x)=0; end
[0138] where "th" represents a threshold level.
[0139] Thus, a pattern D(x) that is divided into regions by 2
values ("0" or "1") is obtained.
[0140] FIG. 9 illustrates an example of a pattern D(x) divided into
N (N=2) regions and FIG. 10 shows frequency characteristics of the
pattern D(x).
[0141] (Process A-2) Process for embedding basic tone patterns
having same characteristics and different phases (the two checkered
patterns 1A, 1B shown in FIG. 1 in this example) in respective
regions in the N-regions division pattern D(x) (refer to FIG.
11).
[0142] Let us call the checkered pattern 1A of FIG. 1 as a
checkered pattern with the "phase 0" and the checkered pattern 1B
as a checkered pattern with the "phase 1" and then classify the N
(N=2) regions-division pattern D(x) shown in FIG. 9 into white and
black regions so as to perform a process of embedding (convolution)
the checkered patterns 1A with the "phase 0" in the white regions
and the checkered patterns 1B with the "phase 1" in the black
regions.
[0143] FIG. 11 is an enlarged view of a part of upper left regions
in FIG. 9. In FIG. 11, the checkered patterns 1A with the "phase 0"
are embedded in the white regions and the checkered patterns 1B
with the "phase 1" are embedded in the black regions. Thus, a first
pattern is created.
[0144] FIG. 12 is a flowchart showing an organized process for
creating a first pattern in the processes A-1 and A-2 described
above. As shown in FIG. 12, the process of creating a first pattern
includes creating a white noise pattern (step S11), performing
convolution of a band-pass filter (step S12) so as to create a
green noise pattern where low-frequency components and
high-frequency components are suppressed and applying N-1 threshold
levels to the green noise pattern to create a pattern divided into
N regions (step S13, refer to FIG. 9).
[0145] The process further includes performing convolution of
different basic tone patterns with respective N regions in the
obtained N-regions division pattern (step S14 of FIG. 12, refer to
FIG. 11).
[0146] The steps S11 to S13 in FIG. 12 correspond to the "process
A-1" and the step S14 of FIG. 12 corresponds to the "process
A-2".
[0147] FIG. 13A is an example of a first pattern created by
processing in the "process A-1" and the "process A-2" and FIG. 13B
is an enlarged view of a part of upper left regions in FIG. 13A.
FIG. 13C shows frequency characteristics (two-dimensional spatial
frequency spectrum) of the pattern shown in FIG. 13A.
[0148] As shown in FIGS. 13A to 13C, the first pattern created
through the "process A-1" and the "process A-2" represents a
pattern in which the neighborhood of each basic tone frequency is
suppressed and the frequency of occurrence in the outer periphery
of each basic tone frequency is increased. The pattern, however,
shows inadequate suppression of low-frequency components as
compared with the desirable pattern (FIG. 2 and FIG. 6).
Accordingly, a process for decreasing low-frequency components is
carried out in the following process B.
[0149] <<Process B (Step S20 in FIG. 8)>>
[0150] The process B (step S20 in FIG. 8) is a process for pattern
optimization of decreasing low-frequency components while
maintaining characteristics of the neighborhood and the outer
periphery of each basic tone frequency, which is obtained in the
process A-2 described above.
[0151] A result of the process will now be described in advance.
FIG. 14A is an example of a pattern (equivalent to a basic pattern)
ultimately created by processing in the process B and FIG. 14B is
an enlarged view of a part of upper left regions in FIG. 14A. FIG.
14C shows frequency characteristics (two-dimensional spatial
frequency spectrum) of the pattern shown in FIG. 14A. FIG. 14A is
on a par with the pattern shown in FIG. 2. Frequency
characteristics of FIG. 14C are equivalent to those of FIG. 6 and
meet the conditions of ideal frequency characteristics described in
FIG. 5.
[0152] Examples of the specific process of the process B will now
be described.
Example 1
Frequency Control Method
[0153] The method of example 1 used for the process B employs
pattern optimization of decreasing low-frequency components while
maintaining pattern characteristics of the neighborhood and the
outer periphery of each basic tone frequency by repeating dot
exchange processing shown in the flowchart of FIG. 15 for a
predetermined number of times. Upon the start of dot exchange
processing shown in FIG. 15, convolution of a dot selection filter
with the first pattern obtained in the process A-2 is performed
(step S22).
[0154] The dot selection filter is a filter created to highlight
low-frequency components and components in the neighborhood of a
basic tone frequency in a pattern. An example of creating a dot
selection filter will be described below. The filter is created by
multiplying a filter for highlighting low-frequency components
(FL), a filter for highlighting components in the neighborhood of a
basic tone frequency (FS) and a transition filter (FT) having a
gradient from low frequency to a basic tone frequency.
[0155] FIG. 16 is an explanatory drawing (two-dimensional) of
frequency characteristics of a dot selection filter. FIG. 17 shows
filter intensity (cross section of the white straight line of FIG.
16) of a dot selection filter in a radial direction from the
original point to the upper right corner (the direction of the
white straight line of FIG. 16) in a two-dimensional frequency
space of FIG. 16. In FIG. 17, the horizontal axis presents
frequency in a radial direction (expressed as "radial frequency")
and the vertical axis presents filter intensity in a logarithmic
scale.
[0156] As shown in FIG. 17, the dot selection filter F is the
product of the low-frequency filter FL, the basic tone frequency
filter FS and the transition filter FT.
[0157] The low-frequency filter is intended for differentiation of
components visually identified, thus it is preferable that the
filter highlight components at frequencies smaller than or equal to
a cut-off frequency which depends on visibility. A filter
formulated by, for example, the following expression 1 can be used
for the filter that highlights low-frequency components
(low-frequency filter FL).
F L ( k ) = ( L 0 - 1 ) ( 1 - k k lcf ) n .theta. ( k lcf - k ) + 1
[ Expression 1 ] ##EQU00001##
[0158] Where the frequency k in a radial direction satisfies
k.gtoreq.0, the cut-off frequency klcf for low-frequency components
is roughly determined by human visibility, and .theta.(x) is a step
function.
[0159] The fitting parameters Lo and n can take on any real numbers
although it is preferable that Lo>1 and n.gtoreq.2 are
satisfied, respectively.
[0160] A filter formulated by, for example, the following
expression 2 can be used for the filter that highlights components
in the neighborhood of each basic tone frequency (basic tone
frequency filter FS).
F S ( k ) = s [ ( L s - 1 ) ( 1 - k s - k k scf ) n s .theta. ( k
scf - k s - k ) + 1 ] [ Expression 2 ] ##EQU00002##
[0161] Where the direct product of s which indicates a plurality of
basic tone frequencies is worked out.
[0162] It is preferable that the cut-off frequency kscl of the
periphery of components at each basic tone frequency be smaller
than or equal to the frequency of the band-pass filter described
above.
[0163] The following filter obtained by computing the Fourier
transform of a Point Spread Function (PSF) which corresponds to a
filter with the inverse of a checkered pattern is used for the
transition filter FT when checkered patterns, for example, are
employed.
FT=abs(fft(PSF))
[0164] where PSF is, for example, one shown in FIG. 18.
[0165] Such a transition filter FT causes components suppressed by
the low-frequency filter FL to transition to band components
(|BPFmax-BPFmin| width component) at the periphery of components at
each basic tone frequency.
[0166] The dot selection filter F is the product of these three
different filters (FL, FS, and FT) and formulated by the following
expression.
F(k)=F.sub.L(k)F.sub.S(k)F.sub.T(k) [Expression 3]
[0167] Performing convolution of the dot selection filter F with a
pattern created in the "process A" provides highlighted
low-frequency components and highlighted basic tone value
components (components at each basic tone frequency and in the
neighborhood thereof expressed as "basic tone value components") in
the pattern.
[0168] Now, let us proceed to step S24 in the flowchart of FIG.
15.
[0169] In the step S24, dot exchange processing in relation to
density is performed. That is, in the pattern obtained by computing
the convolution of the dot selection filter in the step S22, high
density dots and low density non-dots are exchanged (step S24).
Such processing creates a new pattern in which dot arrangement is
changed, which, in many cases, allows low-frequency components and
basic tone value components to be suppressed. This new pattern is
equivalent to a "second pattern".
[0170] Next, an evaluation value on the new pattern created in the
step S24 is calculated (step S26). The dot exchange processing of
the step S24 does not always result in the suppression of
low-frequency components and basic tone value components although
they do in many cases. Thus, the step S26 provides calculation of
an evaluation value to check the effect of suppressing
low-frequency components and basic tone value components by the dot
exchange processing (step S24).
[0171] This evaluation value uses an index correlated with
"decreasing low-frequency components while maintaining
characteristics, obtained in the process A, of the neighborhood of
each basic tone frequency". For example, perform convolution of the
dot selection filter described in FIG. 17 with the pattern obtained
in the step S24 and a standard deviation of density levels in the
pattern (the following [Expression 4]), as one example, can be used
for the evaluation value.
Eval=stdev[iff(F(k)fft(img))] [Expression 4]
[0172] Where Eval is an evaluation function for defining an
evaluation value and the stdev function determines a standard
deviation of a population.
[0173] Following the calculation of an evaluation value in the step
S26, the process determines whether or not the evaluation value has
been improved (step S28). If the evaluation value shows improvement
(smaller value) after the exchange of dots in the step S24 as
compared with that of the pre-exchange, the process determines that
there is an improvement.
[0174] In the case of confirming an improvement in the step S28,
the pattern (dot arrangement) is updated to the new pattern
obtained in the step S24 (step S30).
[0175] If no improvement is confirmed in the step S28, the pattern
is not updated (step S32).
[0176] Repeating the process of FIG. 15 enables the updating of the
pattern with a further improved evaluation value and consequently
the pattern to be optimized
Example 2
Boundary Region Optimization Method
[0177] In the example 1, the dot selection filter includes a basic
tone frequency filter so as to suppress components at each basic
tone frequency. The example 2 achieves the suppression of
components at each basic tone frequency by putting a constraint in
the process of exchanging dots in a real space.
[0178] FIG. 19 is a flowchart showing dot exchange processing
applied in the example 2. In the flowchart of FIG. 19, a step
identical or similar to that of the flowchart in FIG. 15 is
assigned with the same step number of FIG. 15. As shown in FIG. 19,
the flow of the basic process is similar to that of FIG. 15
(Example 1).
[0179] The filter formulated by the expression 3 may be used for a
dot selection filter applied in the step S22 of FIG. 19 as in FIG.
15 (Example 1). It is, however, preferable that the example 2 use
the following expression 5 because components at basic tone
frequencies are suppressed in a real space.
F(k)=F.sub.L(k)F.sub.T(k) [Expression 5]
[0180] In the step S24A of FIG. 19, a "constraint" is placed in the
exchange of dots, and a new pattern where high density dots and low
density non-dots are exchanged is created so as to satisfy the
constraint. In this method, a constraint involving the suppression
of components at basic tone frequencies is added.
[0181] Specifically, in a real space, the clusters of checkered
pattern show the most noticeable pattern characteristics of
components at basic tone frequencies. Thus, dot exchange processing
is performed as in the step S24 of FIG. 15 (Example 1) under a
constraint that "dot arrangement should not be changed inside the
boundaries of clusters of checkered patterns but changed only on
the boundaries" so as to create a new pattern in which
low-frequency components are decreased while the suppression of
components at basic tone frequencies are maintained (step S24A of
FIG. 19).
[0182] FIG. 20 is an enlarged view of clusters of checkered
patterns and their boundaries in a first pattern, illustrated in an
easy-to-understand manner. In FIG. 20, pattern regions shown in
black represent checkered pattern clusters 22 and regions shown in
gray represent boundaries 24 of the checkered pattern clusters
22.
[0183] With reference to FIG. 20, under the constraint that pixel
regions (boundary regions) including the boundaries 24 shown in
gray are extracted from the first pattern and dots are exchanged
only in the boundary regions (gray), a new pattern is created (step
S24A of FIG. 19).
[0184] Subsequent processing for calculating an evaluation value
(step S26), determining improvement in the evaluation value (step
S28), updating the pattern based on the determination (step S30) or
not updating (step S32) is in like manner with that of the example
1 (FIG. 15).
[0185] Processing in the process B as exemplified in FIGS. 15 and
20 causes dot arrangement in the first pattern to be corrected,
providing a basic pattern having target frequency characteristics
(FIG. 6).
[0186] FIG. 21 shows diagrams illustrating an organized arrangement
of patterns in a "real space" and intensity patterns in a "virtual
space" (frequency space) to summarize the processing from (process
A-1) through (process A-2) to (process B) detailed in FIGS. 8 to
20.
[0187] In FIG. 21, patterns obtained in (process A-1), patterns
obtained in (process A-2) and patterns obtained in (process B) are
shown in the upper row, middle-row, and lower-row, respectively.
Patterns in a real space are shown on the left-hand side of FIG. 21
and the intensity distribution in a "virtual space" (frequency
space) is shown on the right-hand side. In FIG. 21, the diagram on
the left in the upper row, the diagram on the right in the upper
row, the diagram on the left in the middle row, the diagram on the
right in the middle row, the diagram on the left in the lower row
and the diagram on the right in the lower row are equivalent to
FIG. 11, FIG. 10, FIG. 13B, FIG. 13C, FIG. 14B and FIG. 14C,
respectively.
[0188] Through the process as shown in FIG. 21, a pattern having
target frequency characteristics can be obtained.
[0189] <Convolution of Basic Tone Patterns with Respective N
Regions in an N-Regions Division Pattern>
[0190] The reason why frequency characteristics obtained in the
process A-1 are deployed in the periphery of each basic tone
frequency in the process A-2, with respect to the description
between the process A-1 and the process A-2, will now be described
in further detail.
[0191] An image divided into N regions (N=2 in this example) in
(process A-1) is defined as Gi(x). That is, the image on the left
in the upper row of FIG. 21 is Gi(x). The subscript i represents a
distinction of N different regions in an N-regions division pattern
and takes natural numbers between 1 and N. For example, when N=2,
G1 with i=1 represents the black part in a pattern and G2 with i=2
represents the white part in the pattern.
[0192] A convolution image, which is embedded in Gi(x) divided into
N regions, is defined as pi(x). In this example, the two checkered
patterns (staggered image) 1A, 1B having different phases described
in FIG. 1 correspond to pi(x).
[0193] A pattern image obtained in (process A-2), that is, the
image shown on the left in the middle row of FIG. 21 is defined as
I(x), so that I(x) with i={1,2} (distinction of N different
regions) is expressed as follows:
I ( x ) = i G i ( x ) p i ( x ) [ Expression 6 ] ##EQU00003##
[0194] A Fourier transform is performed on this (that is, to be
projected in a virtual space) to obtain the following
expression.
F [ I ] = k ' i G i ( k - k ' ) p i ( k ' ) [ Expression 7 ]
##EQU00004##
[0195] Since pi(k) shows the local maximum at each basic tone
frequency and has little components in the area other than basic
tone frequencies, pi(k) can be approximated as follows using the
delta function .delta.(x).
p i ( k ) .apprxeq. s a s .delta. ( k - s ) [ Expression 8 ]
##EQU00005##
[0196] Where s is a basic tone frequency. Substitution of
[Expression 8] into [Expression 7] yields the following expression
F[I], i.e. the FFT image shown on the right in the middle-row of
FIG. 21.
F [ I ] = k ' l G i ( k - k ' ) p i ( k ' ) .apprxeq. k ' i G i ( k
- k ' ) s a s .delta. ( k ' - s ) = s i a s G i ( k - s ' ) [
Expression 9 ] ##EQU00006##
[0197] Incidentally, the image on the left in the upper row of FIG.
21 is expressed as follows:
F [ G ] = F [ i G i ( x ) ] = i G i ( k ) [ Expression 10 ]
##EQU00007##
[0198] When [Expression 9] and [Expression 10] are compared, it is
understood that [Expression 9] has characteristics of [Expression
10] shifted by [s']. In other words, there is a shift of the basic
tone frequency [s'] in characteristics between the image on the
right in the middle-row of FIG. 21 formulated by [Expression 9] and
the image on the right in the upper-row of FIG. 21 formulated by
[Expression 10]. This demonstrates that the frequency
characteristics of the neighborhood of the original point in the
image on the right in the upper-row of FIG. 21 is deployed in the
neighborhood of each basic tone frequency (four corners) in the
image on the right in the middle-row.
[0199] <Configuration of an Apparatus for Creating a Pattern
According to the Embodiment>
[0200] FIG. 22 is a block diagram illustrating a configuration
example of an apparatus capable of creating a pattern (basic
pattern) having the pattern characteristics described in FIGS. 5
and 6.
[0201] A pattern creating apparatus 30 includes a first pattern
creation unit 32 and a pattern optimization processing unit 34 (a
processing unit for suppressing components at basic tone
frequencies and low frequencies) for changing the pattern of a
first pattern created by the first pattern creation unit 32 so as
to suppress components at basic tone frequencies and low
frequencies in the first pattern.
[0202] The first pattern creation unit 32 includes an N-regions
division pattern creation part 40, a basic tone pattern convolution
processing part 50 and a basic tone pattern storing part 52. The
N-regions division pattern creation part 40 includes a white noise
pattern creation part 42, a band-pass filter (BPF) processing part
44, an N-regions dividing part 46 and an N-regions division
threshold storing part 48. The processing unit 34 for suppressing
components at basic tone frequencies and low frequencies includes a
filter processing part 60, a dot exchange processing part 62, an
evaluation value calculator 64, and an updating determination part
66.
[0203] The first pattern creation unit 32 performs processing for
creating a first pattern as described in the step S10 of FIG. 8.
The processing unit 34 for suppressing components at basic tone
frequencies and low frequencies performs optimization of a pattern
as described in the step S20 of FIG. 8.
[0204] The N-regions division pattern creation part 40 of FIG. 22
performs processing described in the steps S11 to S13 of FIG. 12.
The white noise pattern creation part 42 of FIG. 22 performs
processing for creating a white noise pattern as described in the
step S11 of
[0205] FIG. 12. The band-pass filter processing part 44 of FIG. 22
performs processing described in the step S12 of FIG. 12. The
N-regions division threshold storing part 48 of FIG. 22 stores N-1
threshold levels described in the step S13 of FIG. 12. The
N-regions dividing part 46 of FIG. 22 performs processing described
in the step S13 of FIG. 12 by applying N-1 threshold levels stored
in the N-regions division threshold storing part 48.
[0206] The basic tone pattern storing part 52 of FIG. 22 stores N
different basic patterns. The basic tone pattern convolution
processing part 50 performs convolution of different basic tone
patterns with respective N regions in an N-regions division pattern
created by the N-regions dividing part 46 so as to perform
processing described in the step S14 of FIG. 12. The filter
processing part 60 of FIG. 22 performs filter processing described
in the step S22 of either FIG. 15 or FIG. 19. The dot exchange
processing part 62 of FIG. 22 performs processing described in the
step S24 of FIG. 15 or in the step S24A of FIG. 19. The evaluation
value calculator 64 of FIG. 22 calculates an evaluation value as
described in the step S26 of either FIG. 15 or FIG. 19. The
updating determination part 66 of FIG. 22 performs processing for
updating a pattern as described in the steps S28 to S32 of either
FIG. 15 or FIG. 19.
[0207] The processing unit 34 for suppressing components at basic
tone frequencies and low frequencies repeats the process described
either in FIG. 15 or FIG. 19 a plurality of times to optimize the
pattern so that a basic pattern 70 is created (refer to FIG. 22).
Thus, the basic pattern 70 created ultimately has the
characteristics described in FIGS. 5 and 6.
<Quantization Processing>
[0208] Through the undergoing of the processes A and B described
above, a pattern having desired characteristics (FIG. 2 as an
example) for one gray level (equivalent to a record ratio of 50%)
is created. The pattern is designated as a "basic pattern". To
convert (quantize) a continuous-tone image to a binary or
multi-level dot image, the "basic pattern" described above is used.
To make a multi-level quantized pattern (dot image), the amounts of
droplets corresponding to respective multiple levels may vary with
the type of a recording medium for use in printing (paper
type).
Concrete Example 1 of Quantization
Threshold Matrix Method
[0209] In the first example of quantization, a threshold matrix
(also called a "dither matrix") is used. A threshold arrangement is
determined by taking the above-mentioned basic pattern as a dot
arrangement candidate.
[0210] Japanese Patent No. 4143560 discloses a method for making a
threshold matrix by taking a binary image having values
corresponding to a halftone ratio of 50% as a dot arrangement
candidate. A threshold matrix may be made based on the basic
pattern 70 using the method described in Japanese Patent No.
4143560.
[0211] <Method for Determining a Pattern with a Record Ratio
Between 0 and 50%>
[0212] First, a method for making a dot image corresponding to a
gradation range with a record ratio between 0% and 50%.
Determination of a dot arrangement corresponding to each gray level
of an input image enables the creation of a threshold matrix which
achieves such a dot arrangement at each gray level, thus
determining a dot arrangement (dot image) corresponding to each
gray level value and making a threshold matrix are practically
identical.
[0213] <<Method for Determining a Pattern Corresponding to
the Minimum Gray Level Value>>
[0214] First, freely arrange dots as many as the number
corresponding to the minimum gray level value in the dot-on
portions (dot-on pixel portions shown in black in FIG. 2,
hereinafter referred to as "on portions") of a basic pattern
created through the above-mentioned process A, B.
[0215] The minimum gray level value described here is the smallest
gray level value L1 except for "LO" having no dot among discrete
values (L0, L1, . . . Lmax) that can be taken as a gray level value
L in a continuous-tone image. For example, when a gradation is
expressed in 8 bits (0-255), the correspondences can be L0=0, L1=1,
L2=2, . . . , Lmax=255. In the case of a matrix size with p rows
ans q columns and 256 gray levels, the number of dots corresponding
to the minimum gray level value Lmin=1 can be, for example, defined
as pxq/256.
[0216] After that, perform dot exchange processing only in the on
portions of the basic pattern. The dot exchange processing is the
same as the process described in FIGS. 15 and 19. The difference is
that candidates that undergo the dot exchange processing are under
the constraint of the basic pattern (the constraint that dots can
be arranged only in the on portions of the basic pattern).
[0217] It is preferable that the filter of [Expression 5] be used
for the dot selection filter in like manner with the step S 22 of
FIG. 19. Parameters for use may be changed from those for the
creation of the basic pattern.
[0218] Repeat the dot exchange processing to create a pattern
corresponding to the minimum gray level value.
[0219] <<Method for Determining a Pattern Corresponding to a
Value of One Plus the Minimum Gray Level>>
[0220] A method for creating a pattern corresponding to L2 of "+1
gray level" with respect to the minimum gray level value L1:
Arrange a predetermined number of new dots in the on portions of
the basic pattern in addition to the dots already arranged in the
dot arrangement corresponding to the minimum gray level value.
[0221] The "predetermined number" is the number of dots determined
with an intention that how many dots need to be increased to
achieve a density increase in relation to a "+1" increase in gray
level.
[0222] As a simplest example, in the case of a matrix size with p
rows ans q columns and 256 gray levels, the number of dots
corresponding to a "+1" increase in gray level can be, for example,
defined as pxq/256.
[0223] Then, repeat the dot exchange processing in like manner with
FIGS. 15 and 19 only on the newly arranged dots so as to optimize
the dot pattern corresponding to the gray level.
[0224] Repeat the same processing on the subsequent gray levels L3,
L4 . . . in sequence so that optimized patterns corresponding to
the respective gray levels can be created. This provides optimized
patterns corresponding to respective gray levels lower than or
equal to a gray level Ls which correspond to the basic pattern with
a record ratio of 50%.
[0225] <Method for Determining a Pattern with a Record Ratio of
Greater than 50% and Up to 100%>
[0226] To create a dot pattern corresponding to gray levels with a
record ratio of more than 50%, perform the same method as that for
a record ratio between 0 and 50% under the constraint that dots are
added in the off portions of the basic pattern.
[0227] This provides patterns corresponding to all the respective
gray levels.
[0228] Relationships determined between patterns and their
respective gray levels permit the acquisition of a threshold
matrix.
[0229] A comparison of the threshold matrix created above with the
continuous-tone image (input image) enables quantization.
[0230] FIG. 23 is a block diagram illustrating a configuration
example of an image processing apparatus capable of processing
quantization by the use of a threshold matrix. An image processing
apparatus 80 includes an image data input part 82 (equivalent to an
"image input part"), a threshold matrix storing part 84, a
quantization processing part 86 and a quantization image output
part 88. The image data input part 82 is an image input part for
capturing a continuous-tone image 92 (equivalent to "first image
data") which is subject to processing.
[0231] The image data input part 82 acts as a data acquisition part
for capturing image data. The image data input part 82 can be
composed of a data input terminal that captures image data from
another signal processing part outside or inside the apparatus. The
image data input part 82 may be equipped with a wired or wireless
communication interface, may be equipped with a media interface for
reading and writing on an external storage medium (removable disk)
such as a memory card or may be an appropriate combination of these
modes.
[0232] The threshold matrix storing part 84 stores threshold matrix
data acquired by the use of a basic pattern. The quantization
processing part 86 refers to threshold matrix data stored in the
threshold matrix storing part 84 so as to quantize an input
image.
[0233] The quantization image output part 88 outputs a dot image 94
(equivalent to "second image data") of a quantized pattern created
by the quantization processing part 86.
[0234] The quantization image output part 88 can be a communication
interface, a media interface for reading on an external storage
medium such as a memory card, an output terminal for image signals
or another among a variety of forms.
Concrete Example 2 of Quantization
Error Diffusion Method
[0235] In another method of quantization, an error diffusion method
is applied with the use of a basic pattern as a constraint for dot
arrangement.
[0236] For example, for a specific gray level or below, elements
corresponding to "off" portions of the basic pattern are set to off
and only elements corresponding to "on portions" of the basic
pattern are determined to off or on by an error diffusion method.
Also, for a specific gray level or above, elements corresponding to
"on portions" of the basic pattern are definitely set to "on" and
elements corresponding to "off" portions of the basic pattern are
determined to off or on by an error diffusion method. The "specific
gray level" referred to here can be a gray level corresponding to
the basic pattern.
[0237] FIG. 24 is a block diagram illustrating a configuration of
an image processing apparatus capable of processing quantization by
an error diffusion method with the use of a basic pattern as a
constraint.
[0238] In an image processing apparatus 100 shown in FIG. 24,
components identical or similar to those of FIG. 23 are assigned
with the same reference numerals, and their descriptions are
omitted.
[0239] The image processing apparatus 100 of FIG. 24 includes an
image data input part 82; a processing pixel specifying part 104
that specifies pixels (processing pixels) subject to quantization
processing in an input image data; a calculator 106 that adds up
the gray level of a processing pixel and a cumulative value of
diffused quantization errors on the peripheral quantized pixels
(expressed as "gray level+peripheral error calculator" in FIG. 24);
a basic pattern storing part 108 that stores data on a basic
pattern used as a constraint for dot arrangement; a threshold
storing part 110 that stores threshold levels for determining
quantization; a quantization determining part 112 that compares a
value obtained by the calculator 106 with the threshold levels for
determining quantization (also called "thresholds for error
diffusion") and quantizes pixel values so as to meet the constraint
of the basic pattern; a quantization error calculation/diffusion
part 114 that calculates quantization errors to diffuse them into
not-yet-processed pixels in the periphery of a processing pixel; a
processing result memory part 116 that stores quantization results;
and a quantization image output part 88 that outputs quantized
images produced by quantization processing. The quantization
determining part 112 acts as a "quantization processor".
[0240] FIG. 25 shows an example of an error diffusion matrix used
by the quantization error calculation/diffusion part 114. In FIG.
25, "x" represents the position of a pixel subject to quantization
and arrows represent an order in which quantization is processed.
Quantization errors are distributed to each of four
not-yet-processed pixels (right, lower right, immediately
underneath, lower left) adjacent to a pixel of interest
(quantization target pixel x). Error diffusion matrix elements A to
D that specify an error distribution ratio can be set to
appropriate values. When diffused uniformly, for example, errors
are divided into 4 equal parts, thus the distribution ratio is 1/4
each.
[0241] The threshold matrix described in the example 1 may be used
as a basic pattern used as a constraint for quantization by the
error diffusion method.
Concrete Example 3 of Quantization
Threshold Matrix and Error Diffusion Jointly Used Method
[0242] FIG. 26 shows a flowchart of quantization processing in
which a threshold matrix and an error diffusion method are jointly
used. In FIG. 26, dither[x][y] represents a value (elements) at a
position (x, y) in a threshold matrix. th_dth[i][level] represents
a threshold-dependent value for comparison with the threshold
matrix (a threshold for comparison with the threshold matrix)
(i=0,1,2). th_edf[level] represents an error diffusion threshold.
dot[j][level] corresponds to one of dot sizes {no droplet, small
drop, middle drop, large drop} depending on the gray level
(j=0,1,2,3).
[0243] At the start of quantization processing at each pixel shown
in FIG. 26, the process adds up the original gray level of a target
pixel and peripheral errors diffused from the target pixel by error
diffusion so as to calculate a gray level including the peripheral
errors (step S101).
[0244] The process compares the value in the threshold matrix
(dither[x][y]) with the threshold th_dth[i][level] to divide the
image into regions. The threshold th_dth[i][level] is defined for
every gray level of a target pixel and stored in a predetermined
memory beforehand. In this example, the image is divided into 4
regions using a first threshold th_dth[0][level], a second
threshold th_dth[1][level] and a third threshold
th_dth[2][level].
[0245] First, the process compares the value in the threshold
matrix with the first threshold th_dth[0][level] (step S 102). When
the result shows that the value in the threshold matrix is smaller,
the dot size specified by dot[0][level] is selected (step
S103).
[0246] If the value in the threshold matrix is greater than or
equal to the first threshold in the step S102, the process compares
the value in the threshold matrix with the second threshold
th_dth[1][level] (step S104). When the result shows that the value
in the threshold matrix is smaller, the dot size specified by
dot[1][level] is selected (step S105).
[0247] If the value in the threshold matrix is greater than or
equal to the second threshold in the step S104, the process further
compares the value in the threshold matrix with the third threshold
th_dth[2][level] (step S106). When the value in the threshold
matrix is smaller than or equal to the third threshold
th_dth[2][level], the process goes to step S107 in which a
comparison between the gray level including the peripheral errors
and the error diffusion threshold th_edf[level] is made (step
S107). The error diffusion threshold th_edf[level] is also defined
for every gray level of a target pixel and stored in a
predetermined memory beforehand. When the result in the step S107
shows that the gray level including the peripheral errors is
smaller than the error diffusion threshold, the dot size specified
by dot[2][level] is selected (step S108).
[0248] If the gray level including the peripheral errors is greater
than or equal to the error diffusion threshold in the step S107,
the dot size specified by dot[3][level] is selected (step S109). As
described above, a region in which the dither threshold is smaller
than or equal to the third threshold (and greater than or equal to
the second threshold) is binarized by an error diffusion
method.
[0249] If the value in the threshold matrix is greater than the
third threshold in the step S106, the dot size specified by
dot[4][level] is selected (step S110).
[0250] A dot size corresponding to each dot[j][level] can be
appropriately defined for each gray level. dot[0][level],
dot[1][level], dot[2][level], dot[3][level], and dot[4][level] for
certain gray levels can be defined, for example, to small drop,
middle drop, no droplet, large drop, and large drop, respectively.
It is essential only that dot[3][level]>dot[2][level] should be
satisfied, and each value is defined so that large dots are placed
for a large quantization error and small dots are placed for a
small quantization error.
[0251] After a dot size is selected for the target pixel as
described above, a quantization error is calculated (step S111).
The quantization error is an error that occurs when the gray level
including the peripheral errors is quantized and a difference
between the gray level including the peripheral pixels and a
quantization threshold. The quantization threshold is each of gray
levels corresponding to respective dot[0][level], dot[1][level],
dot[2][level], dot[3][level], and dot[4][level].
[0252] The calculated quantization error is diffused to the
peripheral pixels according to a predetermined error diffusion
matrix (refer to FIG. 25) (step S112). Subsequently, the process
shifts its target pixel for quantization to an adjacent pixel and
performs the same processing on it so that all the pixels are
quantized.
[0253] The quantization processing described above determines
respective record ratios of different regions having dot[0][level],
dot[1][level] and dot[4][level], respectively, corresponding to the
steps S103, S105 and S110 according to the threshold matrix and
determines record ratios for the other regions by binarization with
the use of the error diffusion method (steps S108 and S109).
Processing quantization in this manner allows record ratios with 4
levels to be uniquely determined gray level by gray level.
[0254] Although this example uses a threshold which depends on the
original gray level of a target pixel for each threshold
th_dth[i][level], a threshold which depends on the gray level
including the peripheral errors may be used.
[0255] FIG. 27 is a block diagram illustrating a configuration of
an image processing apparatus capable of processing quantization by
the combined use of a threshold matrix and error diffusion as
described in FIG. 26. In the configuration of FIG. 27, components
identical or similar to those of FIGS. 23 and 24 are assigned with
the same reference numerals, and their descriptions are
omitted.
[0256] A quantization processing part 122 of an image processing
apparatus 120 in FIG. 27 includes a dither processing part 86A that
performs quantization using a threshold matrix; and a quantization
determining part 112A that performs quantization by applying
thresholds for error diffusion. The image processing apparatus 120
processes quantization according to the flowchart described in FIG.
26.
[0257] As described above, quantization with the use of a basic
pattern provides high frequent occurrence of basic tone patterns
and a mostly uniform-distributed quantized pattern. This allows
both the expansion of the color reproduction region and the
suppression of artifacts.
[0258] <Pattern Creating Apparatus and Image Processing
Apparatuses in the Embodiments>
[0259] Functions of various components in the pattern creating
apparatus 30, the image processing apparatus 80, the image
processing apparatus 100 and the image processing apparatus 120
exemplified in FIGS. 22, 23, 24 and 27, respectively, can be
implemented by hardware such as a computer and an integrated
circuit or software (program) that effects the operation of a
central processing unit (CPU) and the like, or an appropriate
combination of these.
[0260] In other words, functions of various components in the
pattern creating apparatus 30, the image processing apparatuses 80,
100, and 120 or each step of the processes described in FIGS. 8,
12, 15, 19 and 26, according to the embodiments, can be performed
by computers. Programs for letting the computer perform the
processes and the functions described in the embodiments may be
installed in the computer beforehand or may be provided through a
non-transitory recording medium such as a magnetic disk, an optical
disc, a magneto-optical disk, a memory card or another
computer-readable recording medium (a data storage medium) that
stores the programs (or computer-readable codes of the programs).
In addition, instead of the form of providing programs through such
a tangible object i.e. a storage medium, program signals may be
provided as a download service through a communication network like
the Internet.
[0261] <Definition and Constraint for N when an N-Regions
Division Pattern is Created>
[0262] Although the embodiments described above used the checkered
patterns 1A, 1B of FIG. 1 as basic tone patterns for the sake of
simple explanation, embodiments of the present invention are not
limited to the examples.
[0263] Different patterns convoluted (embedded) in the respective
regions of a pattern divided into N regions by the "process A" for
creating a basic pattern show identical densities per unit area and
have a difference in phase (phase of a spatial cycle of dot
arrangement) or/and basic tone frequency, one to the other.
[0264] N of N regions represents the number of different pattern
(basic tone pattern) types used for convolutions. In the case of
the embodiments using the checkered patterns 1A, 1B exemplified in
FIG. 1, N=2 because two different patterns are used.
[0265] For example, other embodiments using the four different
patterns 131A to 131D of FIG. 28 can be supposed. N=4 when all the
four types are used, and N=3 when only three are selected among
these.
[0266] These N (N types) patterns show identical densities per unit
area and have a difference in phase or/and basic tone frequency,
one to the other.
[0267] A density per unit area for all the patterns (1A, 1B, 131A
to 131D) shown in FIGS. 1 and 28 is 50%. Thus, 2 to 6 types can be
freely selected among these 6 types.
[0268] (Basic Tone Frequency)
[0269] The basic tone frequency is a local maximum obtained when a
frequency analysis is performed on a pattern in which different
basic tone patterns are convoluted with the N respective regions of
the pattern. The basic tone patterns are repeatedly arranged until
the size of the convoluted pattern reaches the size of the original
pattern undergoing the convolution. The basic tone pattern
corresponds to a "specific pattern".
Example 1
[0270] FIG. 29 shows an image of a pattern in a real space (a
checkered pattern image in real space) wherein the pattern has an
arrangement of repeating checkered patterns; and FIG. 30 is a fast
Fourier transformed (FFT) image obtained when a fast Fourier
transform (FFT) is performed on a checkered pattern image in a real
space. Since FIG. 30 is an FFT image, each side of the FFT image
has repeatability and thus each side adjoins another identical
image (not shown).
[0271] The component amounts of frequencies in an FFT image are
represented by brightness, and black indicates "0" component while
white indicates a large component. In the FFT image of FIG. 30, the
basic tone frequency of the checkered patterns is shown at the
"white portion" in the upper left corner area enclosed in the
dashed-line circle 15.
Example 2
[0272] FIG. 31 shows a checkered pattern in which a 1- by 2-pixel
unit is the minimum unit for the white portion or the black
portion. FIG. 31 presents a pattern in which the pattern 131A shown
at the most left of FIG. 28 is repeatedly arranged. FIG. 32 is an
FFT image obtained when a fast Fourier transform is performed on
the real space image of FIG. 31. The pattern of FIG. 31 has two
basic tone frequencies shown at the white portions in the
dashed-line circles 25A, 25B of FIG. 32.
Example 3
[0273] When the checkered pattern of the example 1 and the 1- by
2-pixel checkered pattern of the example 2 are combined, there
exists a total of three basic tone frequencies composed of the
basic tone frequency of the checkered pattern of the example 1
(one) and the basic tone frequencies of the example 2 (two).
[0274] <Configuration of Inkjet Recording Apparatus>
[0275] FIG. 33 is a block diagram illustrating an essential
configuration of an inkjet recording apparatus that includes an
image processing apparatus in accordance with an embodiment of the
present invention. An inkjet recording apparatus 150 includes a
recording head 160, a control unit 170 (equivalent to a "control
device") that controls recording performed by the recording head
160, and a paper conveying part 180 (equivalent to a relative
moving device"). Although the recording head 160 described here is
fit for a single color for the sake of simple illustration, the
inkjet recording apparatus 150 includes a plurality of inkjet heads
corresponding to a plurality of respective ink colors.
[0276] The recording head 160, though shown schematically, includes
a plurality of piezoelectric elements 162, each of which acts as a
discharge energy generating element so as to generate energy needed
for discharging ink in relation to each nozzle, and a switch IC 164
for switching between driving/non-driving each of the piezoelectric
elements 162.
[0277] The number of nozzles, the density and the arrangement of
nozzles in the recording head 160 are not particularly limited and
various forms may be adopted. For example, to achieve a designated
recording resolution in a main scanning direction, a
one-dimensional nozzle arrangement in which a large number of
nozzles are arranged in a straight line (single row) at regular
intervals may be used, or a so-called staggered arrangement in
which two nozzle rows are displaced half a pitch the pitch of
nozzles (a nozzle-to-nozzle pitch) in a nozzle row direction with
respect to each other may be used. In addition, to achieve a
further high recording resolution, the arrangement can be, for
example, a matrix in which three or more nozzle rows are arranged,
i.e. a large number of nozzles are two-dimensionally arranged on an
ink discharge surface (nozzle surface).
[0278] In the case of an inkjet head having a two-dimensional
nozzle arrangement, a projection nozzle row in which each nozzle is
projected (orthogonal projection) so as to arrange each nozzle
along the direction of paper width (equivalent to a main scanning
direction) in the two-dimensional nozzle arrangement can be thought
of equivalent to a single nozzle row in which nozzles are arranged
generally at regular intervals in a main scanning direction (medium
width direction) with a nozzle density which allows the achievement
of a recording resolution. The "regular intervals" described here
represent practically regular intervals that produce droplet jetted
points printable in an inkjet printing system. For example, the
concept of "regular intervals" includes a case in which slightly
variant intervals are contained with consideration given to the
movement of droplets on a medium owing to a margin of error or
impact interference. Taking the projection nozzle row (also called
a "practical nozzle row") into consideration allows correspondences
to be established between the order in which projection nozzles are
arranged in a main scanning direction and the positions of nozzles
(nozzle numbers). The "nozzle position", if any, described later
refers to the position of the nozzle in the practical nozzle
row.
[0279] The control unit 170 includes a system control part 171, an
image data input part 172 that acts as an input interface receiving
the original image data of an image to be recorded (multi-level
image data), and an image processing part 174 that performs density
correction and quantization on an input image data. The control
unit 170 further includes a driving waveform generator 176 and a
head driver 178.
[0280] The image processing part 174 is a signal processing device
which converts an input image data to binary or multi-level dot
data (quantization data). The image processing apparatuses 80, 100,
and 120 can be applied to the image processing part 174.
[0281] A mode with the threshold matrix described in FIG. 23, a
mode with the error diffusion method described in FIG. 24 and a
mode with the combined use of the threshold matrix and error
diffusion described in FIGS. 26 and 27 can be used as a
quantization processing (halftone processing) device.
[0282] The quantization processing generally converts m-level (m:
an integer of 3 or greater) image data to n-level (n: an integer of
2 or greater and less than m) image data that has a smaller gray
level than m. While the simplest example is conversion into binary
dot image data (dots on/off), the quantization processing is also
capable of quantizing multi-level values corresponding to different
dot sizes (for example, three sizes, i.e. large dots, medium dots,
small dots).
[0283] Binary or multi-level image data (dot data) generated in the
image processing part 174 is used as ink discharge control data
(droplet jetting control data) for controlling driving
(on)/non-driving (off) of each nozzle and further droplet amount
(dot size) in the case of multi-level values. The dot data (droplet
jetting control data) generated in the image processing part 174 is
fed to the head driver 178 that controls the discharge of ink from
the recording head 160.
[0284] The driving waveform generator 176 is a device which
generates voltage signal waveforms for driving the piezoelectric
elements 162 corresponding to respective nozzles of the recording
head 160. Driving voltage signal waveform data is stored in a
storage device such as ROM beforehand and waveform data for use is
output as needed. Signals (waveforms) generated by the driving
waveform generator 176 are supplied to the head driver 178. Signals
output from the driving waveform generator 176 may be digital
waveform data or analog voltage signals.
[0285] The inkjet recording apparatus 150 shown in this example
employs a driving method for supplying a common driving voltage
waveform signal via the switch IC 164 to each of the piezoelectric
elements 162 of the recording head 160 and switching on/off of each
corresponding switch element connected to each electrode of each of
the piezoelectric elements 162 in response to the discharge timing
of each of the respective nozzles so as to cause the nozzle
corresponding to each of the piezoelectric elements 162 to
discharge ink. The recording head 160 discharges a droplet of ink
on demand from a driving signal and a discharge control signal
given from the head driver 178.
[0286] The system control part 171, the image data input part 172
and the image processing part 174 in FIG. 33 constitute the
equivalent of an "image processing apparatus".
[0287] The system control part 171 controls the paper conveying
part 180 through a paper conveying control part 182. This causes
paper sheets (recording medium, not shown) to be conveyed to the
recording head 160. The paper conveying control part 182 and the
paper conveying part 180 are equivalent to a "relative moving
device".
Advantages of Embodiments
[0288] According to the embodiments of the present invention
described above, components in the neighborhood of each basic tone
frequency are suppressed and a local maximum occurs in the outer
periphery of each basic tone frequency. This permits the
acquisition of a pattern that presents high frequent occurrence of
basic tone patterns and decreased low-frequency components in the
clusters of the basic tone patterns and in the whole pattern. This
allows both the expansion of the color reproduction region and the
suppression of artifacts.
[0289] <Device for Relatively Moving Head and Paper>
[0290] Embodiments of the present invention are not limited to the
configuration in which the recording medium is conveyed with
respect to a stopped recording head as in the embodiments
exemplified above, but can use a configuration in which the
recording head moves with respect to a stopped recording medium.
Although line heads in a single-path system are normally arranged
in a direction perpendicular to a recording medium feeding
(conveying) direction, a mode in which recording heads are arranged
in a direction slanting by a given angle with respect to the
direction perpendicular to the conveying direction may be
adopted.
[0291] The present invention can also be applied to an image
forming apparatus in a serial scanning system in which image
recording is performed with the recording head scanning in a
direction perpendicular to the recording medium conveying direction
without limiting its application to a single-path system.
[0292] <Recording Media>
[0293] The term "recording medium" is a general term for media on
which dots are recorded with a recording head and includes terms
such as recording medium, printing medium, recorded medium,
image-formed medium, receiving medium and discharged medium. The
present invention can be applied to various media without
limitation in material, form and the like, including continuous
roll paper, cut paper, seal sheets, resin sheets and other OHP
sheets, films, cloth, nonwoven fabrics, printed circuit boards on
which a wiring pattern is formed, and rubber sheets.
[0294] <Applicable Apparatuses>
[0295] The above-described embodiments have been discussed with
examples in which the present invention is applied to an inkjet
recording apparatus for graphic printing. The applicability of the
present invention, however, is not limited to the examples. The
present invention can be applied to a wide variety of inkjet
apparatuses that draw various forms and patterns with the use of
functional liquid materials, including apparatuses that draw wiring
patterns on electronic circuits, apparatuses for manufacturing
various devices, resist recording apparatuses that use resin liquid
as a functional liquid for discharge, color filter manufacturing
apparatuses, and apparatuses for forming microstructures with the
use of deposition materials.
[0296] <Recording Head Usage Modes Other than Inkjet
System>
[0297] Although the embodiments in the above explanation show an
inkjet recording apparatus as an example of an image forming
apparatus using a recording head, the applicability of the present
invention is not limited to this. In addition to an inkjet system,
the present invention can be applied to image forming apparatuses
in various systems that use dot recording, including thermal
transfer recording apparatuses equipped with a recording head
having thermal elements as recording elements, LED
electrophotographic printers equipped with a recording head having
LED elements as recording elements and silver halide photographic
printers having a LED exposure line head.
[0298] It should be understood that appropriate structural
modifications, additions and deletions may occur in the embodiments
of the present invention described above insofar as they are within
the scope of the present invention. Application of the present
invention is not limited to the embodiments described above, and it
should be understood by those skilled in the art that various
alterations may occur insofar as they are within the scope of the
present invention.
* * * * *